A High Performance Server for Data-Intensive Computation

麻省理工学院的计算和系统生物学计划（CSBi）要求资金购买具有存储区域网络数据存储（SMP-SAN）的对称多处理服务器，使科学家能够处理与各种系统生物学项目相关的数据密集型计算问题。这种集中访问的计算资源是从事生物研究基本方面的几个实验室立即取得进展所迫切需要的。在这个建议中，我们描述了一个重要的和不断增长的子集的计算生物学问题，由于大的数据大小不能使用传统的32位集群连接到网络附加存储（NAS）设备上的数据来解决。我们的目的是增加CSBi高性能计算（HPC）核心和Whitehead-MIT生物成像中心中可用的计算资源，其中SMP服务器的基本配置包含16个64位Itanium 2处理器，具有64千兆字节（GB）内存，运行通用的64位Linux操作系统（OS），以及数TB的数据存储，具有多个光纤通道连接，可实现每秒数千兆位的低延迟带宽。SMP-SAN将允许研究人员轻松收集数百GB到TB的数据，并直接解决本提案中描述的两个主要生物学领域的计算需求：成像和蛋白质组学。这些项目包括开发成像技术，研究细胞行为和癌症诊断的非侵入性方法，识别，理解和建模蛋白质，以帮助药物发现和临床诊断。该资源将位于Whitehead-MIT生物成像中心，并与CSBi HPC资源集成。生物成像中心和CSBi HPC Core在高性能计算系统和大型数据存储系统的管理方面拥有丰富的经验，并致力于支持和维护拟议的系统。

DESCRIPTION (provided by applicant): The Computational and Systems Biology Initiative at MIT (CSBi) requests funds to purchase a symmetric multi-processing server with storage area network data storage (SMP-SAN) enabling scientists to handle data-intensive computational problems related to a wide variety of Systems Biology projects. This centrally- accessible computational resource is critically required for the immediate progress of several labs engaged in fundamental aspects of biological research. In this proposal, we describe a significant and growing subset of computational biology problems that due to large data sizes cannot be tackled using traditional 32-bit clusters connected to data on network attached storage (NAS) devices. It is our intention to augment the computational resources available in the CSBi High-Performance Computing (HPC) core and the Whitehead-MIT Biolmaging Center with a base configuration of an SMP server containing sixteen 64-bit Itanium2 processors with 64 gigabytes (GB) memory, running a generic 64-bit capable linux operating system (OS), and several terabytes (TBs) of data storage with multiple fiber-channel connections enabling multi-gigabit per second low latency bandwidth. The SMP-SAN will allow researchers to easily collect hundreds of GB to TB of data and directly address the computational needs in two major areas of biology described in this proposal: imaging and proteomics. The projects range from development of imaging technologies to study cellular behavior and non-invasive methods for cancer diagnosis to identifying, understanding and modeling proteins to help in drug discovery and clinical diagnosis. The resource will be located in the Whitehead-MIT Biolmaging Center and integrated with the CSBi HPC resources. The Biolmaging Center and the CSBi HPC Core have extensive experience with management of high-performance computing systems and large data storage systems and are committed to the support and maintenance of the proposed system.